<br><br><div class="gmail_quote">On Wed, Oct 13, 2010 at 2:41 AM, David van der Spoel <span dir="ltr"><<a href="mailto:spoel@xray.bmc.uu.se">spoel@xray.bmc.uu.se</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div class="im">On 2010-10-13 01.54, <a href="mailto:chris.neale@utoronto.ca" target="_blank">chris.neale@utoronto.ca</a> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
I have 28,000 .xtc files, each having a single frame and each 150K. If I<br>
run du -hs on the directory containing these .xtcs, I get 4.4 GB.<br>
Nevertheless, when I run trjcat -f *.xtc -o ../tot.xtc , my memory<br>
consumption goes over 11 GB and then I run out of available memory.<br>
</blockquote></div>
The program reads the first frame of all files, and since we roughly have three times compression from xtc to float your figures of 4 vs. 11 Gb. match perfect.<br></blockquote><div>Each first frame is read twice. Thus I think it is 2x the memory. </div>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<br>
Although this is an extreme case, where there is a clear workaround (more memory) it could be worthwhile looking into an enhancement that releases the memory again after checking the first frame of all files. It is anyway impossible to keep 28,000 files open. Please file a bugzilla as an enhancement.<br>
</blockquote><div>Yes and it should be easy because the frame isn't reused. Thus freeing it shouldn't have any side effect.</div><div><br></div><div>Roland</div></div>